Google Street View is an extraordinarily expensive project for a company which normally deals with razor-slim margins. It involves building customised cars, shipping them all over the world, and then hiring drivers to patrol the roads for hours on end.
The eventual plan is to map every street they can (and they mean every – Jon Rafman’s 9-eyes is a wonderful collection of weirder pictures taken), an extraordinary project which certainly goes far beyond what makes economic sense. While Street View images of, for example, London’s Oxford Street are likely to be regularly checked and probably easily monetiseable, it’s hard to imagine what use images of Manitoba, Canada’s highway 39 are, beyond bragging rights for the company.
Now, I’m realizing the biggest Street View data coup of all: those vehicles are gathering the ultimate training set for driverless cars.
I’m sure this is obvious to people who have followed it more closely, but the realization has really blown my mind. With the goal of photographing and mapping every street in the world, Street View cars must encounter every possible road situation, sort of by definition. The more situations the driverless car knows about, the better the training data, the better the machine-learning algorithms can perform, the more likely it is that the driverless car will work. Brilliant.
Google is, first and foremost, a company build around data-wrangling. Most of the data they get is provided by their users, but some, like the Street View corpus, they have to go out and get. And if they do, it’s worth their while to work out as many ways of using that data as possible. The real question is whether they realised once they had all the information that they could use it to teach computers how to drive, or if this has been their cunning plan all along.
Thanks to Robin Sloan for the pointer.